8649

8030

كارشناسي ارشد

كنترل

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1392

ده،82ص.: مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

جعفر قيصري

يدا... ذاكري

16/1/93

جواد عسكري، مرضيه كمالي

مهندسي برق و كامپيوتر

ID8030

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

Model Predictive Controller Design in Industrial Automation Networked Control Systems Farid Alavi f alavi@ec iut ac ir Date of Submission 12 10 2013 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language FarsiSupervisor J Ghaisari ghaisari@cc iut ac irAdvisor Y Zakeri yzakeri@cc iut ac irAbstractIn Networked Control Systems NCSs all or some parts of control systems such as sensors controllers and actuators communicate together via digital networks Due to non idealcommunication environment data packet losses and delays are unavoidable Thus anycontroller design algorithm should consider these flaws Model Predictive Control MPC isone of the most popular control methods in industrial process especially in chemical oil andpetrochemical industries Since MPCs usually are employed in slow rate processes NCS canprovide effective and reasonable platform to implement control systems In this thesis a newapproach is proposed to design MPC controllers in networked control systems First a modelis introduced for the control system such that data packet losses and delays in network areincluded in the design procedure To increase the accuracy of the model data exchangeprotocol is considered based on IEEE 802 15 4 which is a network protocol for industrialautomation applications The proposed design method is derived from infinite horizon MPCs At each sampling time the controller solves an optimization problem with some constraintsexpressed as LMIs Simulation results indicate that the total elapsed time to compute futurecontrol commands in each sampling time is sufficiently small to be applicable in real timeimplementations of MPCs To conclude feasibility it is also proved that any LMI created inthe controller design has a real feasible solution Furthermore input and output constraints areconsidered in the design process Finally the proposed design procedure is extended toachieve a robust model predictive controller design method in NCS In this case it is assumedthat the plant uncertainty can be expressed as polytopic Then it is proved that the controllerdesign procedure guarantees asymptotical stability conditions for the equilibrium point of thesystem A simulation for the proposed robust controller shows that in presence of modeluncertainties and time delays in the network data exchange the system remains stable Keywords Model Predictive Control MPC Networked Control System NCS LinearMatrix Inequality Polytopic Uncertainty Compensate Delay and Packet Loss LyapunovFunction

جعفر قيصري

يدا... ذاكري

جواد عسكري، مرضيه كمالي